LLaMA-33B TPS calculator

Open weights Meta AI 32.5B parameters February 2023

Each card below is assessed against this model at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from the card's memory bandwidth and the size of the model once compressed.

Calculated for this model

132 cards that can run it

818 cards we hold specifications for

Smallest card that fits

RTX A4500

20 GB · Q3_K_M · 22.5 tok/s

Fastest card

B200

104 tok/s · 180 GB

Which GPUs can run LLaMA-33B?

Set the inputs, read the answer

Context length tokens

A longer conversation needs more memory, which can push this model off smaller cards.

Minimum quality

Hides cards that would only fit the model by compressing it below this point.

Search cards

Card memory

Also show cards that cannot run it

132 cards match

Calculating

					Needs	Quantisation	Fit
104 tok/s 89–125	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	36.5 GB	Q8_0	Comfortable
104 tok/s 89–125	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	36.5 GB	Q8_0	Comfortable
83.3 tok/s 50–133 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	36.5 GB	Q8_0	Comfortable
83.3 tok/s 50–133 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	36.5 GB	Q8_0	Comfortable
66.6 tok/s 40–107 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	36.5 GB	Q8_0	Comfortable
63.7 tok/s 54–76	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	36.5 GB	Q8_0	Comfortable
63.7 tok/s 54–76	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	36.5 GB	Q8_0	Comfortable
61.0 tok/s 37–98 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	36.5 GB	Q8_0	Comfortable
54.1 tok/s 32–87 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	36.5 GB	Q8_0	Comfortable
54.1 tok/s 32–87 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	36.5 GB	Q8_0	Comfortable
54.1 tok/s 32–87 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	36.5 GB	Q8_0	Comfortable
51.3 tok/s 44–62	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	36.5 GB	Q8_0	Comfortable
43.8 tok/s 37–53	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	36.5 GB	Q8_0	Comfortable
43.8 tok/s 37–53	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	36.5 GB	Q8_0	Comfortable
43.8 tok/s 37–53	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	36.5 GB	Q8_0	Comfortable
43.8 tok/s 37–53	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	36.5 GB	Q8_0	Comfortable
43.8 tok/s 37–53	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	36.5 GB	Q8_0	Comfortable
43.5 tok/s 37–52	DRIVE A100 PROD NVIDIA	32 GB	1,870 GB/s	May 2020	25.2 GB	Q5_K_M	Tight
43.5 tok/s 37–52	GRID A100A NVIDIA	32 GB	1,870 GB/s	May 2020	25.2 GB	Q5_K_M	Tight
41.7 tok/s 35–50	GeForce RTX 5090 NVIDIA	32 GB	1,790 GB/s	Jan 2025	25.2 GB	Q5_K_M	Tight
41.7 tok/s 35–50	GeForce RTX 5090 D NVIDIA	32 GB	1,790 GB/s	Jan 2025	25.2 GB	Q5_K_M	Tight
40.3 tok/s 34–48	GeForce RTX 5090 D V2 NVIDIA	24 GB	1,340 GB/s	Aug 2025	21.4 GB	Q4_K_M	Tight
36.7 tok/s 31–44	A30X NVIDIA	24 GB	1,220 GB/s	Apr 2021	21.4 GB	Q4_K_M	Tight
33.3 tok/s 20–53 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	36.5 GB	Q8_0	Comfortable
33.3 tok/s 20–53 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	36.5 GB	Q8_0	Comfortable

Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.

On record

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation: Meta AI
Organisation type: Industry
Country: United States of America
Published: 27 February 2023
Authors: Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, Guillaume Lample

What it does

The problem areas the model was built for. A model can carry several of each.

Domain: Language
Task: Language modeling, Code generation, Language modeling/generation, Question answering

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters: 32.5B
Training data: 1,400,000,000,000 tokens
Epochs: 1.04
Batch size: 4,000,000

Training compute

The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.

Training compute: 2.7 × 10²³ FLOP
How it was established: Operation counting

Availability

Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.

Weights: Open — downloadable
Model access: Open weights (non-commercial)
Training code: Unreleased

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Likely above 10²³ FLOP: Yes
Record confidence: Confident
Citations: 19,926
Benchmark data: LLaMA-33B

Sources

Where this record came from and when it was last checked.

Reference: LLaMA: Open and Efficient Foundation Language Models
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run LLaMA-33B

Ranked by estimated tokens per second, newest card first where speeds tie. Because generation is bound by memory bandwidth, this ordering follows bandwidth rather than any gaming benchmark.

The smallest GPUs that still run LLaMA-33B

The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.

What the numbers mean

What you need to run it

Minimum card

RTX A4500

Memory needed

17.6 GB

Fastest

104 tok/s

LLaMA-33B reaches a parameter count of 32.5B. That lands in the range a serious desktop card can handle once the weights are compressed. The number of cards we track that can run it: 132.

The least hardware that works is RTX A4500, with a memory capacity of 20 GB, running it at a compression of Q3_K_M and producing around 22.5 tokens per second.

Top of the range is B200, generating roughly 104 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

Where it came from

LLaMA-33B was published by Meta AI, in the country recorded as United States of America, during February 2023. The category the publisher falls under is industry.

It works in the domain of Language, and is recorded as performing the task of language modeling, Code generation, Language modeling/generation, Question answering.

Published weights mean the model runs on your machine rather than someone else's, which is what makes the hardware question below answerable at all.

Understanding the speeds

Half the cards that hold it manage more than 22.1 tokens per second. Producing text faster than most people read it: 106 of them.

It is a dense model, so every parameter is read for every token produced. That makes speed track memory bandwidth almost exactly — a card with twice the bandwidth generates roughly twice as fast.

Because the architecture is recorded, the memory column is derived rather than estimated.

Training and provenance

The training run consumed about 2.7 × 10²³ FLOP. That figure measures what producing the model cost, and has no bearing on how fast it answers.

It was trained on a corpus of about 1,400,000,000,000 tokens of text.

Step by step

How to choose a GPU for LLaMA-33B

The table above has already assessed every card we hold specifications for against this model. Getting to your answer takes six steps.

01

Start from the memory column

Every card here has been checked against LLaMA-33B, needing around 17.6 GB at a compression of Q3_K_M. No amount of processing power compensates for a card that cannot hold it.
02

Match the context to your actual use

Set the context to what you will actually use. The cache grows with the conversation, and it is the usual reason a card that seemed fine stops fitting LLaMA-33B.
03

Set a quality floor

Compression is what makes a model fit smaller cards, at some cost in accuracy, reaching a compression of Q3_K_M on the smallest card that fits. Setting a minimum quality drops the cards that only manage it by squeezing further than you would want, and holds the comparison at one level.
04

Compare tokens per second, not specifications

Sort by speed to see how cards rank for LLaMA-33B. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 104 tok/s.
05

Look at the headroom, not just the fit

The fit column separates cards that just manage it from those with room to spare, in the case of LLaMA-33B. Comfortable means you can grow the context later. That difference matters more than a few tokens per second, so buy for comfortable if you expect to.
06

See what else that card runs

Following a card through to its own page shows every other model it can hold, which is the question that follows once you have settled on LLaMA-33B.

Answers

LLaMA-33B — common questions

LLaMA-33B— how accurate are these speed estimates?

They are calculated from specifications rather than measured, and each carries a range. One example: 89–125 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

LLaMA-33B— what GPU do I need to run it?

The smallest card in our catalogue that holds it is RTX A4500, with a memory capacity of 20 GB. It runs the model at a compression of Q3_K_M using about 17.6 GB, and produces roughly 22.5 tokens per second. The number of cards able to run it in total: 132.

LLaMA-33B— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 104 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and the number of cards clearing that: 106.

LLaMA-33B— how much VRAM does it need?

It needs about 17.6 GB at a compression of Q3_K_M, which is what the smallest card that runs it uses. Less compression needs more: the figures in the memory column above are recalculated for each card, because each one holds the least-compressed version it can.

LLaMA-33B— can I run it on a GPU holding 24 GB?

Yes. The card GeForce RTX 5090 D V2, holding 24 GB, runs it at a compression of Q4_K_M, using about 21.4 GB and generating roughly 40.3 tokens per second. The fit is tight.

LLaMA-33B— is it open source?

Its weights are published, so it can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.

LLaMA-33B— how many parameters does it have?

It has a parameter count of 32.5B. Table 2 in the paper. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

LLaMA-33B— who created it?

It was published by Meta AI, based in United States of America, an organisation categorised as industry.

LLaMA-33B— when was it released?

It was published in February 2023. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

LLaMA-33B— what is it used for?

It works in the domain of Language, and is recorded as handling the task of language modeling, Code generation, Language modeling/generation, Question answering. These are the areas it was designed around; they describe intent rather than a hard boundary.

LLaMA-33B— where can I download it?

The weights are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

LLaMA-33B— how much compute was used to train it?

Training consumed around 2.7 × 10²³ FLOP. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.

LLaMA-33B— can I run it if it does not fit in my GPU?

Only by offloading, which is usually a false economy: the part held in system memory drags the whole thing down. The nearest miss we calculate falls short by 7.0 GB. Every figure here assumes the whole model is resident on the card.

LLaMA-33B— would two GPUs run it faster?

A second card roughly doubles the memory available but not the generation rate. The number already able to run it alone: 132. So a second card is rarely the answer here.

LLaMA-33B— why does the quantisation differ between cards?

Because capacity varies, so does how hard it has to be squeezed. The number of distinct levels in the table above: 5. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

Source

Original publication

Record last updated 25 May 2026

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.

All AI models